The scientists sizing up climate change's role in a disaster

Once a week, Ben Clarke’s team meets to talk about the weather. Unfortunately, they have no shortage of topics: In mid-September, they discussed Typhoon Dujuan, which struck eastern Japan. The month before, they talked about the flash floods and debris flows in Nepal, likely triggered by the collapse of a glacier and rock wall. After each discussion, they decide whether to analyze the role of climate change in the event.

“We choose based on a combination of how impactful it has been, and to an extent, how much media or public interest there is,” says Clarke, a climate scientist at Imperial College London and a member of World Weather Attribution (WWA), an international team that studies climate change’s impact in extreme weather events. They also may study an event because of its scientific significance.

His team of about 20 work in the field known as extreme event attribution. To inform journalistic coverage, they work at a clip faster than conventional science, finishing and publicizing each study in one to two weeks. They later publish about a quarter of their analyses in peer-reviewed journals. (“The main findings of our rapid studies have never changed after being submitted, reviewed and accepted by an academic journal,” the WWA website reads.)

The social importance of the field is evident. For example, the floods in Nepal killed 1,400 people, with 5,200 reported missing. The economic impact can be staggering, too: In 2025, there were 55 natural disasters that caused billion-dollar damages, according to a report by insurance broker Gallagher Re.

Extreme event attribution is also increasingly shaping law and policy, appearing in reports by the United Nations Intergovernmental Panel on Climate Change and in environment-related litigation in the U.S. For example, in the ongoing Multnomah v. Exxon Mobil Corp., a county government in Oregon has argued that oil companies should pay $51.5 billion for damages incurred from a 2021 heat wave in which 69 people died. (The Supreme Court is expected to rule next year on whether a similar lawsuit, brought by Boulder, Colorado, can move forward, a decision with profound implications for climate litigation.)

Clarke’s team uses a framework known as the “probabilistic approach” to study the events. Imagine two many-sided, loaded dice. Each face of the first die represents a possible daily temperature in New York City in January, and each face of the second die represents a possible daily temperature in July. The January die would be more likely to land on 0°C, whereas the July die would be more likely to land on 30°C.

A key step in Clarke’s workflow is to determine the likelihood the loaded die will land on each side, known as the die’s probability distribution. For example, say the number on each side represents the hottest annual temperature in Karachi, Pakistan, which experienced a heat wave peaking above 46°C this April and May, killing at least 10 people. Using historical data, Clarke’s team calculates the likelihood of the temperature hitting 46°C under both pre-industrial average global temperatures and current average global temperatures. From this, they can determine whether the heat wave’s occurrence correlates with rising average global temperatures.

Then, they assess whether to assign causation to rising global temperatures, rather than merely correlation. To do this, they use multiple climate models to create many simulations of the area under warming climate conditions. If the simulations generally recreate the extreme weather event, the team uses the models for analysis — and if the models show similar trends to historical observations, they have the confidence in actually attributing causality to climate change, says Clarke.

In the case of the 2026 Pakistan heat wave, they found climate change approximately tripled its likelihood. “The same event would have been about 1°C cooler in a preindustrial climate,” they wrote in their report.

Of various frameworks used in the field, the probabilistic method has been in use the longest. The authors of the first paper, published in 2004 to link the prior-year European heat wave to greenhouse gas emissions, used the approach. However, it has its limitations, says Ted Shepherd of the University of Reading in the U.K. The probabilistic approach often simplifies a complex event into a single variable, such as rainfall. Consequently, it is ill-suited to analyze what are known as “compound events,” where multiple extreme events occur together, like flooding from a glacier collapse in Nepal. In addition, humans generally struggle with translating probabilities into practical decisions.

Shepherd thinks another framework is more practical. In the “storyline approach,” researchers place the weather event in a narrative timeline and investigate how various physical factors — say, wind speed, temperature, or pressure — contributed to it, and how the event might unfold differently if these factors changed. A storyline research question might, for example, simulate 2012’s Hurricane Sandy, but with higher sea level. This approach can help answer the question, “If the [extreme weather event] happens again, would we be ready?” says Shepherd, a pioneer of the method.

The storyline approach can also offer more physics-based information than the probabilistic approach, which relies more on statistics, according to Shepherd. Its more concrete descriptions — which lay out the forces and thermodynamic variables at play — can also help clarify the physical sources of uncertainty in the analysis.

Shepherd also finds the storyline approach useful for informing practical decisions in private industry. For example, the storyline framework can help engineers decide certain thresholds, like what wind strength a bridge or building can withstand.

Indeed, each method has its strengths and weaknesses, and climate scientists are developing frameworks beyond the probabilistic and storyline approaches to study extreme weather. The “cutting edge” in the field for “gaining the most understanding of the role of climate change is to use multiple different methods,” says Clarke.

Climate scientists have also begun to adopt techniques known as “rare event sampling,” used in condensed matter and particle physics, says Shepherd. By definition, extreme weather event attribution focuses on unlikely events that still occurred. Methods for searching for rare particles in physics experiments might serve useful for understanding the circumstances that led to, say, an unlikely storm.

The field is still young, with opportunities for physicists to contribute. For example, physicists could help improve the accuracy of simulating jet streams, the fast, narrow current in the atmosphere “that drives weather events,” says Clarke. Its methods encompass much of physicists’ toolkit, including thermodynamics, Newton’s laws, probability, and computational modeling.

However, the Trump administration’s attacks on climate science have added uncertainty to the future of extreme event attribution in the U.S. In July, Trump wrote a post on Truth Social calling to defund the National Academies of Science, Engineering, and Medicine, following the academies’ publication of a report on extreme event attribution.

Still, the societal demand for the research is evident, says Clarke. “I actually think the appetite for reliable, credible scientific information is really high,” he says.